White-matter diffusion anisotropy after chemo-irradiation: a statistical parametric mapping study and histogram analysis.

The aim of the study was to evaluate white-matter (WM) diffusion anisotropy in medulloblastoma survivors after cranial irradiation and chemotherapy using voxel-based analysis with SPM99 and fractional anisotropy (FA) histogram-derived indices, and to identify quantitative indices for detecting and monitoring children with treatment-induced white-matter injury. Familywise error rate (FWE) that corrects for multiple comparisons was used to locate statistically significant regions of P < 0.05 in voxel-based analysis. Subsequently, the false discovery rate (FDR) controlling procedure (corrected P < 0.05) was used. FA map histogram analysis of histogram-derived indices, mean FA, mean FA peak height, and peak location was performed. Two-sample t test was used in all analyses. Using FWE-corrected P < 0.05, there was a cluster of reduced anisotropy in the periventricular white matter lateral to the left ventricular atrium. When FDR-corrected P < 0.05 was used, there were multiple clusters of reduced anisotropy in the periventricular white matter, the corpus callosum, and corona radiata. Simplified voxel-based morphometry (VBM)-like analysis of cerebrospinal fluid (CSF) did not show significant differences between patient and control subjects. 'White-matter FA map' histogram showed significant reduction in mean FA and mean FA peak location and significant increase in mean FA peak height in the patient group compared to control subjects (P = 0.003, P = 0.003, and P = 0.014, respectively). This approach of quantifying FA can be applied to characterize anisotropy in the white matter after cranial irradiation and chemotherapy and can potentially be used to detect and monitor treatment-induced neurotoxicity.